david279 Posted December 7, 2020 Share Posted December 7, 2020 The second one is something i never tried before with a mulitfunction on the 00 bus so it maybe a long shot. 1 Quote Link to comment
derpuma Posted December 7, 2020 Share Posted December 7, 2020 Thx for the help. Will try this and report back! Not sure if I am able to test it right now, for sure tomorrow morning. Quote Link to comment
giganode Posted December 7, 2020 Share Posted December 7, 2020 1 hour ago, derpuma said: Yeah, that is the reset problem. Hoped the fix would have been compiled into the kernel. For that reason I use a script via CA User Scripts Plugin. The use avoids me from restarting the whole server. It simply powers the 5700xt off/on and then you are able to reboot the VM. Be sure not to simply copy the script, enter the individual PCI root for your card. I use VFIO-PCI Config and it shows me the the PCI address of my card. I am familiar with this procedure. I think thats a specific problem on macOS vm as the problem does not occur on linux or windows vms. I am thankful for your idea, but that's just another workaround - which by the way puts the whole system to sleep and requires physical interaction with the server. Quote Link to comment
derpuma Posted December 7, 2020 Share Posted December 7, 2020 55 minutes ago, giganode said: I am familiar with this procedure. I think thats a specific problem on macOS vm as the problem does not occur on linux or windows vms. I am thankful for your idea, but that's just another workaround - which by the way puts the whole system to sleep and requires physical interaction with the server. Hmm, I am not 100% sure but I thought I have to reset the 5700XT also when I shut down my Windows 10 VM and start it again. As said not 100% sure cause 99% of my work I use mac OS. But I will check this out! 1 Quote Link to comment
ghost82 Posted December 7, 2020 Share Posted December 7, 2020 5 minutes ago, derpuma said: Hmm, I am not 100% sure but I thought I have to reset the 5700XT also when I shut down my Windows 10 VM and start it again. I agree with you, the issue is related to linux/amd/passthrough and not to which vm runs. 1 Quote Link to comment
derpuma Posted December 7, 2020 Share Posted December 7, 2020 2 hours ago, david279 said: The second one is something i never tried before with a mulitfunction on the 00 bus so it maybe a long shot. David, thx so far. I tried both methods. Unfortunately the card does not show up in mac OS, neither with multifunction='on' nor with both on bus 00 and slot 09 function 0x0 and 0x1... Could it be another problem? Maybe kext related? Any other idea? Quote Link to comment
david279 Posted December 7, 2020 Share Posted December 7, 2020 6 minutes ago, derpuma said: David, thx so far. I tried both methods. Unfortunately the card does not show up in mac OS, neither with multifunction='on' nor with both on bus 00 and slot 09 function 0x0 and 0x1... Could it be another problem? Maybe kext related? Any other idea? Download ioregexplorer and see if its loaded in there. If you see it then you could add it thru the DeviceProperties/Add section in opencore. 1 Quote Link to comment
derpuma Posted December 7, 2020 Share Posted December 7, 2020 (edited) This is what ./gfxutil -f ethernet gives me. So do I have to set 3 DeviceProperties? One for my en0: <key>Add</key> <dict> <key>PciRoot(0x1)/Pci(0x1,0x0)/Pci(0x0,0x0)/Pci(0x1,0x0)</key> <dict> <key>built-in</key> <data>AQ==</data> </dict> </dict> <key>Delete</key> <dict/> And two for the 2x 10Gbe ports on the 10Gb network card? <key>DeviceProperties</key> <dict> <key>Add</key> <dict> <key>PciRoot(0x1)/Pci(0x9,0x0)</key> <dict> <key>built-in</key> <data>AQ==</data> </dict> </dict> <key>Delete</key> <dict/> </dict> <key>Add</key> <dict> <key>PciRoot(0x1)/Pci(0x9,0x1)</key> <dict> <key>built-in</key> <data>AQ==</data> </dict> </dict> <key>Delete</key> <dict/> Edited December 7, 2020 by derpuma Quote Link to comment
david279 Posted December 7, 2020 Share Posted December 7, 2020 I wouldn't add anything for en0 just the 10Gbe devices. Quote Link to comment
ghost82 Posted December 7, 2020 Share Posted December 7, 2020 (edited) 41 minutes ago, derpuma said: And two for the 2x 10Gbe ports on the 10Gb network card? Take into account that you don't have to write 2 times the DeviceProperties section. Just add the 2 addresses into the "add" section. Not sure, but the en0 should be the builtin, maybe you have to readjust something. Maybe resetting network sertings could reorder the en(x) devices, auto setting en0 for the builtin. Otherwise, after resetting networks just boot with the builtin ethernet only, so that the os can set it to en0. Once it's saved as en0 add the others. Edited December 7, 2020 by ghost82 Quote Link to comment
derpuma Posted December 7, 2020 Share Posted December 7, 2020 Can you give me a code example please?! Quote Link to comment
ghost82 Posted December 7, 2020 Share Posted December 7, 2020 8 minutes ago, derpuma said: Can you give me a code example please?! <key>DeviceProperties</key> <dict> <key>Add</key> <dict> <key>PciRoot(0x1)/Pci(0x9,0x0)</key> <dict> <key>built-in</key> <data>AQ==</data> </dict> <key>PciRoot(0x1)/Pci(0x9,0x1)</key> <dict> <key>built-in</key> <data>AQ==</data> </dict> </dict> <key>Delete</key> <dict/> </dict> Quote Link to comment
giganode Posted December 8, 2020 Share Posted December 8, 2020 13 hours ago, ghost82 said: I agree with you, the issue is related to linux/amd/passthrough and not to which vm runs. I use a patched kernel. So no, its just related to my macOS vm.. everything else works fine. 1 Quote Link to comment
ghost82 Posted December 8, 2020 Share Posted December 8, 2020 2 hours ago, giganode said: I use a patched kernel. So no, its just related to my macOS vm.. everything else works fine. Very interesting, may I ask, for pure information, which patch did you apply to the kernel or from where you downloaded? Quote Link to comment
giganode Posted December 8, 2020 Share Posted December 8, 2020 52 minutes ago, ghost82 said: Very interesting, may I ask, for pure information, which patch did you apply to the kernel or from where you downloaded? I'm currently running 6.9.0-beta35 with the old Navi Reset Patch. You can find it here: https://forum.level1techs.com/t/navi-reset-kernel-patch/147547/47 It's no longer supported, because gnif works on something new. But that new project is in the early stages and cannot supersede the old patch as Navi and Vega cards are not working correctly, atm. I tested it. As soon as the new vendor-reset - https://github.com/gnif/vendor-reset - does work, we should give it a try. And maybe - or better hopefully - the new fix can deal with macOS vms better than the old patch. Quote Link to comment
ich777 Posted December 8, 2020 Share Posted December 8, 2020 1 hour ago, ghost82 said: Very interesting, may I ask, for pure information, which patch did you apply to the kernel or from where you downloaded? I made the implementation to Unraid and @giganode tested it for me because I don't own such a card, if you also want to apply the patch then please write me a PM and I will send you a link and how to apply it. Keep in mind that I implemented this because the vendor patch isn't working properly for now. 1 Quote Link to comment
ghost82 Posted December 8, 2020 Share Posted December 8, 2020 1 minute ago, ich777 said: if you also want to apply the patch then please write me a PM and I will send you a link and how to apply it. Thank you very much, I don't own any amd card too, I asked just for curiosity, it seemed strange that it worked for all (?) vms except the mac os one. Quote Link to comment
giganode Posted December 8, 2020 Share Posted December 8, 2020 3 minutes ago, ghost82 said: Thank you very much, I don't own any amd card too, I asked just for curiosity, it seemed strange that it worked for all (?) vms except the mac os one. ok... 17 hours ago, ghost82 said: I agree with you, the issue is related to linux/amd/passthrough and not to which vm runs. But if you haven't got one... then don't deny my statement... Quote Link to comment
ghost82 Posted December 8, 2020 Share Posted December 8, 2020 1 minute ago, giganode said: then don't deny my statement... I didn't want to offend anyone, this must be clear; since it's a patch applied to linux for vfio it should be independent on which vm is applied. You could have other issues in the config of the bootloader for example, that was all I wanted to say. Quote Link to comment
ich777 Posted December 8, 2020 Share Posted December 8, 2020 9 minutes ago, ghost82 said: it seemed strange that it worked for all (?) vms except the mac os one. This doesn't seem strange to me, think of it as a real hackintosh... Even to get it working on bare metal is a little wonder (sometimes, Clover,...)... Keep in mind if you start a Windows VM in between the reset patch works again I think... Quote Link to comment
ghost82 Posted December 8, 2020 Share Posted December 8, 2020 (edited) 1 minute ago, ich777 said: This doesn't seem strange to me, think of it as a real hackintosh... Even to get it working on bare metal is a little wonder (sometimes, Clover,...)... Keep in mind if you start a Windows VM in between the reset patch works again I think... Yes but this doesn't apply to the patch, but as I wrote it's related to wrong configuration of the bootloader (or not working code) which manages the gpu. Edited December 8, 2020 by ghost82 Quote Link to comment
ich777 Posted December 8, 2020 Share Posted December 8, 2020 Just now, ghost82 said: Yes but this doesn't apply to the patch, but as I wrote it's related to wrong configuration of the bootloader which manages the gpu. Why did you know that? Clover was just an example. Did you know how Apple shuts down a system and what it sends to the hardware and what not, I don't. My guess is that Apple does stop the system differently than a standard system look at the many posts out there about that ACPI is diffent to a standard Windows/Linux machine and that this is the problem that the patch triggers the reset of the card. 1 Quote Link to comment
ich777 Posted December 8, 2020 Share Posted December 8, 2020 10 minutes ago, ghost82 said: 🏳️ ? I also don't want to offend anyone but keep in mind I also don't know how Apple shuts down a system and that what I've wrote above is just my oppinion what it can be that the patch doesn't work or is triggered correctly. 1 Quote Link to comment
ghost82 Posted December 8, 2020 Share Posted December 8, 2020 (edited) Me too, I don't know, that was the meaning of my white flag. Sometimes the issue with the english language makes my thoughts misrepresented. I didn't want to write that giganode was not right and I was right, I don't know in details how mac os works, but I have some experience with mac os vm and bootloaders and I'm a curious person in general, that's why I asked even if I don't own an amd gpu. So I'm not offended at all Edited December 8, 2020 by ghost82 Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.